The field of robotics is witnessing significant developments in manipulation and control, with a focus on safety, adaptability, and generalization. Researchers are exploring innovative approaches to improve robotic interaction with complex environments, including the use of vision-language models, force-impedance control, and soft robotics. These advancements have the potential to enable robots to operate effectively in human-centered environments, with improved success rates and reduced force violations. Noteworthy papers include: OmniVIC, which presents a self-improving variable impedance controller with vision-language in-context learning for safe robotic manipulation, achieving a significant increase in success rates. SoftMimic, which introduces a framework for learning compliant whole-body control policies for humanoid robots from example motions, enabling robots to respond compliantly to external forces while maintaining balance and posture.